CN111050659B

CN111050659B - Fetal movement detection method based on Doppler ultrasound signals

Info

Publication number: CN111050659B
Application number: CN201880054310.8A
Authority: CN
Inventors: 张瑛
Original assignee: Koninklijke Philips NV
Current assignee: Koninklijke Philips NV
Priority date: 2017-08-23
Filing date: 2018-08-09
Publication date: 2023-06-02
Anticipated expiration: 2038-08-09
Also published as: BR112020003490A2; EP3672493B1; WO2019038097A1; RU2762818C2; CN111050659A; JP2020526356A; EP3672493A1; RU2020111708A; JP6938763B2; RU2020111708A3

Abstract

A method (100) of detecting fetal movement includes: deriving a plurality of peaks from a doppler ultrasound signal acquired from a subject, each peak of the plurality of peaks being associated with an envelope of a signal segment of the doppler ultrasound signal; calculating a density of the peaks from a plurality of amplitude grids, the density indicating a number of peaks in each amplitude grid having an amplitude; selecting a fetal movement threshold based on the calculated density; and determining whether the signal segment of the doppler ultrasound includes fetal movement by comparing the amplitude of a peak associated with the envelope of the signal segment to the fetal movement threshold.

Description

Fetal movement detection method based on Doppler ultrasound signals

Technical Field

The following generally relates to the Doppler ultrasound arts, the fetal monitoring arts, the fetal movement monitoring arts, and related arts.

Background

For example, it is known to use fetal monitors employing Doppler ultrasound to detect fetal movement, such as "Automated Detection of Fetal Movements in Doppler Ultrasound Signals versus Maternal Perception" by Wrobel et al (journal of medical information and technology, volume 23, pages 43-50, 2014). In the method of Wrobel et al, the Doppler ultrasound amplitude signal is first low pass filtered over time to identify and remove high frequency components attributable to the fetal heart beat. The resulting low-pass filtered signal is referred to as a continuous activity map because it is continuous and is expected to capture mainly fetal movement. The data is processed in segments of one second, and each segment is marked as moving if the number of samples exceeding the threshold exceeds a limit. This converts the continuous activity map into a binary activity map.

Wrobel et al employ an actively adjusted threshold generated as follows. The threshold is initially set to a lower value. Thereafter, each 1 second segment was processed to identify the average of the last 25% of the one second segments. The threshold is adjusted to half the average unless the threshold is below the initial minimum threshold, in which case the latter is selected. The method of Wrobel et al has some difficulty in threshold selection. It assumes that a certain distribution of clinical data may limit the application of the method, as it may or may not be suitable for a particular patient.

CN103190913 provides a fetal movement identification method, which comprises: a fetal movement curve is acquired, a fetal movement threshold is set according to the fetal movement curve, and whether the fetus moves is identified according to the magnitude values of points on the fetal movement curve and the fetal movement threshold.

Accordingly, improved systems and methods for fetal movement detection are desired.

Disclosure of Invention

In one disclosed aspect, a method of detecting fetal movement is disclosed. A plurality of peaks is derived from doppler ultrasound signals acquired from the fetus. The density of the peaks is calculated from the peak amplitudes. A fetal movement threshold is selected based on the calculated density of the peaks. Determining whether the Doppler ultrasound signal segment being measured includes fetal movement by comparing the amplitude of a peak derived from the Doppler ultrasound signal segment being measured with the fetal movement threshold.

In another disclosed aspect, an apparatus for detecting fetal movement includes a doppler ultrasound apparatus comprising: a Doppler ultrasound transducer for acquiring Doppler ultrasound signals from a fetus; and at least one electronic processor programmed to: a plurality of peaks is derived from the doppler ultrasound signal, a density of the peaks is calculated from peak amplitudes, and a fetal movement threshold is selected based on the calculated density of the peaks. In some embodiments, the fetal movement threshold is selected as the peak amplitude at which the density of peaks is greatest. For example, peak amplitudes with the greatest peak densities may be identified by binning a plurality of peaks into peak amplitude bins and identifying the peak amplitudes of the peak amplitude bins that contain the greatest number of peaks. In some embodiments, the at least one electronic processor is further programmed to: the Doppler ultrasound signal is segmented to identify a measured Doppler ultrasound signal segment, and a determination is made as to whether the measured Doppler ultrasound signal segment includes fetal movement by comparing the amplitude of a peak derived from the measured Doppler ultrasound signal segment to the fetal movement threshold.

One advantage resides in determining true fetal movement.

Another advantage resides in improved differentiation of real fetal movement from non-real fetal movement.

Another advantage resides in providing a fetal movement threshold for detecting fetal movement that is tuned for a fetus being tested.

Another advantage resides in providing one or more of the above benefits with improved computational efficiency.

Another advantage resides in automatically providing one or more of the foregoing benefits without relying on the subjective perception of fetal movement by the mother.

A given embodiment may provide zero, one, two, more, or all of the foregoing advantages, and/or may provide other advantages that will become apparent to one of ordinary skill in the art upon reading and understanding the present disclosure.

Drawings

The disclosure may take form in various components and arrangements of components, and in various operations and arrangements of operations. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention.

FIG. 1 illustrates exemplary flowchart operations of a fetal movement detection method;

FIG. 2 schematically shows a more detailed illustrative embodiment of the process of FIG. 1;

fig. 3 schematically shows a parallel graph illustrating the fetal movement threshold selection operations of fig. 1 and 2 for illustration purposes;

figure 4 shows an illustrative doppler ultrasound device suitable for use in performing fetal movement detection of figures 1 or 2.

Detailed Description

Improved methods for adjusting thresholds for fetal movement detection in fetal monitors employing doppler ultrasound are disclosed below. The threshold is automatically determined by the doppler ultrasound system without human intervention. For this purpose, a plurality of peaks is calculated from the Doppler ultrasound signal. The peak density is determined from the peak amplitudes, and then the fetal movement threshold is selected as the peak amplitude with the highest calculated density. The movement is identified as a doppler ultrasound signal peak whose amplitude exceeds a selected movement threshold.

Advantageously, the fetal movement threshold is determined without reference to any input from the mother. In contrast, in some other methods, the mother provides a training label, for example, by pressing a button when the mother feels the fetal movement. While such data is considered additional training information or information for verifying fetal movement thresholds, the disclosed method for selecting fetal movement thresholds does not rely on such input from the mother. This is beneficial because the mother's perception of fetal movement is subjective and can sometimes be erroneous.

Referring to fig. 1, a flow chart schematically shows an illustrative embodiment of a fetal movement detection method 100. At 102, doppler ultrasound signals are acquired. For example, the Doppler ultrasound signal is sampled at a predetermined sampling rate or sampling period (e.g., samples are acquired every 25 milliseconds). During the ultrasound signal acquisition 102, an ultrasound transducer is placed over the maternal abdominal region, the ultrasound transducer being positioned to sonicate the infant in the uterus. In some examples, the doppler ultrasound signal can be filtered using, for example, a low pass filter. The main purpose of the low-pass filtering is to remove the fetal heart beat signal component (and optionally the maternal heart beat signal component); accordingly, the frequency cutoff of the low pass filter should be below the lowest trusted fetal pulse frequency (and optionally below the lowest trusted maternal pulse frequency), in some non-limiting embodiments, the frequency is on the order of 60-90Hz, for example.

At 104, an envelope is calculated for each signal segment of the acquired Doppler ultrasound signal. Typically, for processing convenience, the Doppler ultrasound signal is divided into time slices (sometimes also referred to as time slices). In some embodiments, segmentation also allows for adaptive threshold adjustment. In some examples, the doppler ultrasound signal may be segmented into a plurality of signal segments, which may optionally partially overlap. In some non-limiting illustrative examples, each segment has a duration of 2 seconds with 2000 samples/segment. The envelope calculation for each segment may for example be performed using signal rectification and then smoothing using a low pass filter, but more generally any envelope detector may be used. It is envisioned that the low pass filtering is fused with an envelope detector to remove the heart signal.

At 106, a plurality of peaks are derived from the calculated envelope of the Doppler ultrasound signal segment. For example, each peak is a local maximum detected in the calculated envelope. Each peak of the plurality of peaks is associated with a signal segment of the doppler ultrasound signal that contains the peak.

At 108, a density of the derived peaks is determined from the peak amplitudes. In an illustrative method, the density of the derived peaks is calculated from a plurality of amplitude grids or bins. The calculated density can then be quantized to the number of peaks with amplitude in each amplitude grid or bin.

At 110, a fetal movement threshold is selected based on the calculated density. In one example, the fetal movement threshold is selected as a grid or bin of amplitude peak amplitude values with maximum density values.

At 112, a fetal movement threshold is used for fetal monitoring. To this end, measured signal segments of the Doppler ultrasound signal from the fetus being monitored are filtered and enveloped, similar to

operations

102, 104, and peaks are detected, similar to operation 106, and it is determined whether fetal movement is included in the measured signal segments by comparing peaks associated with the envelopes of the measured signal segments with a fetal movement threshold determined at 110. At 112, if the peak in the measured signal segment exceeds the fetal movement threshold, it is determined that fetal movement is included in the measured signal segment.

Referring now to FIG. 2, a more detailed illustrative embodiment of the process of FIG. 1 is described. In this illustrative example, the most recent data segment or data segment of the Doppler ultrasound signal is acquired every 0.25 seconds at operation 202. Although a 0.25 second segment is processed in the illustrative fig. 2, more generally, the duration of the segment should be short enough to provide near real-time operation while being large enough to capture the fetal movement segment. At operation 204, the segment is preprocessed with a low pass filter to extract the signal component that contains primarily fetal movement information. The original Doppler ultrasound signal (FIG. G1) and the low-pass filtered signal (FIG. G2) are shown on the left side of FIG. 2. In these graphs G1, G2, the x-axis plots one point generated every 0.25s. The sampling rate of these points along the x-axis is also 0.25s. A 2s (with 2000 samples) data segment of the doppler ultrasound signal is input into the calculation. The sampling rate of the Doppler ultrasound signal is 1 millisecond/sample. This is merely an illustrative example, and other fragment sizes and sampling rates may be employed. In operation 206, an envelope is calculated for the signal segments of the filtered data. At 208, a maximum peak is determined for each signal segment associated with the envelope. The curve G3 on the left side of fig. 2 shows an illustrative example of the calculated envelope and the peak marked by a cross, in this example the peak amplitude is about 240. In FIG. G3, the range is 0-200, since downsampling is to a sampling rate of 1/10. At 208, the peak amplitude is added to a peak buffer. An illustrative peak buffer denoted BF is shown schematically on the left side of fig. 2 and is a buffer having a fixed length of time of 100 (which contains a history of peak amplitudes for time segments of every 0.25 seconds), as a non-limiting example.

With continued reference to fig. 2 and with further reference to fig. 3, in operation 210 shown in fig. 2, a fetal movement threshold is calculated or updated based on the peaks in the peak buffer BF. Typically, the fetal movement threshold is determined as the peak amplitude with the highest peak density observed in the peak buffer BF. For illustration purposes, the selection of fetal movement thresholds is schematically illustrated using a parallel graph G4 shown in fig. 3. A simplified illustrative example of a parallel graph is shown as graphical example G5, the graphical example G5 plotting data for only three peaks. In the illustrative parallel graph, peak amplitude data for each Doppler ultrasound signal segment envelope is plotted on parallel axes, with one parallel axis representing time and the other parallel axis representing peak amplitude. This can best be seen in the simplified graph G5, which simplified graph G5 shows only three peaks at times T1, T2, T3, which have corresponding peak amplitudes A1, A2, A3, respectively. As best seen in the parallel graph G4 of data showing the entire peak buffer BF (e.g., 100 time slices in the illustrative example), the peak amplitude axis exhibits the region of highest density (i.e., having the largest number of peak amplitudes). As shown in fig. G4, the peak amplitude with the highest peak density is selected as the fetal movement threshold. Empirically, it has been found that using the peak amplitude at which the maximum number of peaks is observed as the fetal movement threshold is the best choice for the fetal movement threshold, since using such a threshold choice provided marks the peaks as fetal movement or fetal non-movement more consistent with basic real information, e.g. indicated by the mother, than higher or lower thresholds. More advantageously, such threshold selections can be derived in real time from the Doppler ultrasound signal itself during fetal monitoring, as disclosed herein.

Fig. 3 provides a conceptual illustration of selecting a fetal movement threshold with a peak amplitude of the highest peak density in the peak buffer BF. Operation 210 of fig. 2 is typically not implemented using a parallel graph, but rather operation 210 of fig. 2 is implemented algorithmically using data such as a peak buffer BF suitably stored as data points TS, AMP, where TS represents the timestamp of the peak and AMP represents the magnitude of the peak. In one approach, the data points are binned into different amplitude grids or bins, and the amplitude grid or bin with the highest data point count (i.e., the highest number of peaks falling in the peak amplitude bin) is selected as the threshold. The binning width is selected to provide the required resolution for the fetal movement threshold (a smaller binning width provides a higher resolution) while being wide enough to suppress noise due to a limited number of data points.

This is merely one illustrative algorithmic method for determining the peak magnitude of the highest peak density, and other algorithmic methods are also contemplated. For example, in another approach, a Kernel Density Estimate (KDE) of peak density versus peak amplitude may be calculated, and the fetal movement threshold is selected as the peak amplitude at which the KDE exhibits its maximum value. In the KDE method, each data point (i.e., peak) is represented by a gaussian kernel (or other selected kernel function) centered on the peak amplitude and having a selected variance along the peak amplitude axis, and these gaussian values are summed and normalized to generate the KDE.

With continued reference to fig. 2 and 3, at 212, the fetal movement threshold selected at 210 is used to classify peaks due to fetal movement or peaks not due to fetal movement. As indicated schematically in the parallel graph G4 of fig. 3, any peaks of the measured signal segments having an amplitude greater than the selected fetal movement threshold are counted as fetal movements, while the measured signal segments having an amplitude below the selected fetal movement threshold are discarded.

Advantageously, the fetal movement threshold may optionally be dynamically adjusted for the particular fetus being monitored. In this method, the Doppler ultrasound signal 202 is a portion of the Doppler ultrasound signal acquired for the fetus being monitored and the fetal movement threshold is adaptively tuned by repeating

steps

204, 206, 208, 210 while Doppler ultrasound data is being collected. To begin the threshold adjustment process, a hard-coded default fetal movement threshold may be initially used and adjusted (i.e., adjusted) over time for the particular fetus being monitored. In one adaptive approach, once a sufficient time interval for the Doppler ultrasound signal to be acquired by repeating

steps

204, 206, 208 for a succession of acquired Doppler ultrasound signal segments (e.g., sufficient to fill the peak buffer BF), a fetal movement threshold is determined by running step 210 that is applied to the full peak buffer BF. Thereafter, the fetal movement threshold output by step 210 may be fixed for future fetal movement detection. Alternatively, the threshold can be dynamically determined based on a greater number of samples or updated samples and made more accurate over time. For example, the peak buffer BF may be considered a first-in-first-out (FIFO) buffer such that when newer peak data is added to the buffer BF, the oldest peak data will be discarded, and operation 210 may occasionally be applied to the current contents of the peak buffer BF to adaptively update the fetal movement threshold.

In another contemplated non-adaptive approach, training Doppler ultrasound signal data may be acquired from a population of fetal patients that are expected to represent a typical fetus. The training Doppler ultrasound signal data is used as the ultrasound signal 202 to which

operations

204, 206, 208, 210 are applied to generate fetal movement thresholds. This "trained" threshold is then hard coded into the software of the Doppler ultrasound device for use in operation 212, which in this embodiment is performed on the fetus being clinically monitored (which is typically not part of the training population) to evaluate movement of the fetus being clinically monitored.

Referring to fig. 4, an illustrative doppler fetal monitoring device or system 10 configured to detect fetal movement using the methods of fig. 1 or 2 is shown. As shown in fig. 4, the device 10 includes a doppler ultrasound device 12 having an ultrasound transducer 14. Figure 4 shows three transducers 14 arranged on a tray or container 15 so as to be conveniently used with the doppler ultrasound device 12 in an obstetrical ward or other medical environment where multiple patients are likely. The ultrasound transducer 14 is configured to acquire Doppler ultrasound signals. For example, the ultrasound device 12 can be securedOr otherwise attached to the abdominal region of a mother (not shown) carrying a fetus (not shown) such that the ultrasound transducer 14 covers portions of the fetus (e.g., the fetus and the transducer separated by the abdominal region of the mother). Some non-limiting illustrative examples of suitable Doppler ultrasound devices include Philips Avalon with associated wired or wireless ultrasound transducer attachments ^TM A series of fetal monitors.

The Doppler ultrasound device 12 also includes typical components, such as at least one electronic processor 22 (e.g., a microprocessor or microcontroller and auxiliary electronics including internal components indicated in phantom in the illustrative embodiment of FIG. 4), various user-operable controls 24 for performing setup and control of the monitoring session, and a display device or component 26. In some embodiments, the display device 26 can be a separate component from the Doppler ultrasound device 12 having a wired or wireless operational connection. The display device 26 is capable of displaying ultrasound images and also displaying information regarding fetal movement detected at operation 112 of fig. 1 or at operation 212 of fig. 2 for fetal movement monitoring purposes. Fetal movement information may be displayed in various ways, such as a count of fetal movements over a defined time interval, which may scroll over time (e.g., display the number of fetal movements detected in the last three minutes).

The at least one electronic processor 22 is operably connected to a non-transitory storage medium (not shown) that stores instructions that are readable and executable by the at least one electronic processor 22 to perform a fetal movement detection method or process 30, including peak density-based adaptive adjustment of fetal movement thresholds as disclosed herein (e.g., using the method of fig. 1 or 2). The non-transitory storage medium may include, for example, a hard disk drive or other magnetic storage medium; a solid state drive, flash drive, electronically erasable programmable read-only memory (EEPROM), or other electronic memory that stores firmware for the doppler ultrasound device 12; optical discs or other optical storage devices; various combinations thereof, and the like. The non-transitory storage medium may store one or more modules to perform the operations shown in fig. 1 and 2. For example, the non-transitory storage medium can include: a filtering module 32 programmed to filter the doppler ultrasound signals (as described in operations 102 and 204); an envelope and peak calculation module 34 programmed to calculate an envelope from the filtered doppler ultrasound signal (as described in operations 104 and 206) and to determine a peak in the doppler ultrasound signal from the envelope (as described in operation 106); a peak buffer update module 36 programmed (as described in operation 208) to add the determined maximum peak value to the peak buffer; a calculation module 38 programmed (as described in operation 108) to calculate a density of peaks from the derived magnitudes of peaks; a determination module 40 programmed (as described in operations 110 and 210) to determine a dynamic fetal movement threshold from the calculated density; and a classification module 42 programmed (as described in operations 112 and 212) to classify the peak as "moving" or "no moving" based on the determined fetal movement threshold. Although in the illustrative example the on-board processor 22 of the Doppler ultrasound device 12 performs the fetal movement detection method or process 100 of FIG. 1, in other contemplated embodiments the fetal movement detection method or process 100 of FIG. 1 may be performed by cloud processing running on a processor operatively connected via the Internet and/or a hospital data network or the like. Additionally, the methods 100 and/or 200 may be implemented by software (e.g., a computer program or non-transitory computer program product), hardware (e.g., a hard disk drive or other magnetic storage medium, a solid state drive, a flash drive, an electrically erasable programmable read-only memory (EEPROM) or other electronic memory storing firmware, an optical disk or other optical storage device, various combinations thereof), or combinations thereof.

The present disclosure has been described with reference to the preferred embodiments. Modifications and alterations will occur to others upon reading and understanding the preceding detailed description. It is intended that the disclosure be construed as including all such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. A method of detecting fetal movement, comprising:

deriving a plurality of peaks from a doppler ultrasound signal acquired from a fetus, each peak of the plurality of peaks being associated with an envelope of one of a plurality of signal segments of the doppler ultrasound signal;

calculating the density of the peak value according to the peak value amplitude;

selecting a fetal movement threshold based on the calculated density of peaks; and is also provided with

Determining whether the Doppler ultrasound signal segment under test includes fetal movement by comparing the amplitude of a peak derived from the envelope of the Doppler ultrasound signal segment under test with the fetal movement threshold.

2. The method of claim 1, further comprising:

the Doppler ultrasound signal is segmented into the plurality of signal segments.

3. The method of claim 2, wherein at least two adjacent signal segments of the plurality of signal segments overlap in time.

4. A method according to claim 2 or 3, wherein:

the calculating includes calculating the density of the peaks from a plurality of peak amplitude grids, the density being indicative of a number of peaks in each amplitude grid; and is also provided with

The selecting includes selecting the fetal movement threshold as a peak amplitude in the peak amplitude grid at which the calculated density of the peaks is greatest.

5. The method of claim 1, wherein the doppler ultrasound signal segment being measured is acquired from the same fetus as the doppler ultrasound signal from which the plurality of peaks were derived.

6. The method of claim 4, wherein the deriving, calculating, and selecting are iterated to adaptively tune the fetal movement threshold for the fetus.

7. The method of claim 4, further comprising:

acquiring the Doppler ultrasound signals from which the plurality of peaks from the fetus are derived;

storing the derived plurality of peaks in a peak buffer, including updating the contents of the peak buffer using first-in first-out updates during said acquiring of the Doppler ultrasound signals;

wherein the deriving, the calculating, and the selecting are repeated as the contents of the peak buffer are updated to adaptively tune the fetal movement threshold for the fetus.

8. An apparatus (10) for detecting fetal movement, the apparatus comprising:

a Doppler ultrasound transducer (14) for acquiring Doppler ultrasound signals from a fetus; and

at least one electronic processor (22) programmed to:

deriving (34) a plurality of peaks from the Doppler ultrasound signal, each peak of the plurality of peaks being associated with an envelope of one of a plurality of signal segments of the Doppler ultrasound signal;

-calculating (38) the density of the peaks from the peak amplitudes;

selecting (40) a fetal movement threshold based on the calculated density of peaks; and is also provided with

Determining (42) whether the measured Doppler ultrasound signal segment includes fetal movement by comparing the amplitude of a peak derived from the envelope of the measured Doppler ultrasound signal segment with the fetal movement threshold.

9. The device of claim 8, wherein the at least one electronic processor is further programmed to: the Doppler ultrasound signal is segmented into the plurality of signal segments.

10. The apparatus of claim 9, wherein at least two adjacent signal segments of the plurality of signal segments overlap in time.

11. The device of claim 9, wherein the at least one electronic processor (22) is further programmed to:

calculating the density of the peaks from a plurality of peak amplitude grids, the density being indicative of the number of peaks in each amplitude grid; and is also provided with

The fetal movement threshold is selected as the peak amplitude of the peak with the greatest density calculated in the peak amplitude grid.

12. The device of claim 11, wherein the at least one electronic processor is further programmed to: storing the derived plurality of peaks in a peak buffer, including updating the contents of the peak buffer using first-in first-out updates during said acquiring of the Doppler ultrasound signals; and repeating the dividing, the deriving, the calculating, and the selecting as the content of the peak buffer is updated to adaptively tune the fetal movement threshold for the fetus.

13. A non-transitory computer readable medium having program code stored thereon, the program code being readable and executable by one or more electronic processors to perform operations comprising: